After having been distant by the viral success of Chatgpt, the giant of Mountain View now imposes itself in the generative AI, with assets in the infrastructure that he alone can boast.
For a long time to laggle from Openai, Google has operated a real update in AI: unification of its R&D, pierced with the general public with Gemini, technological demonstration with Veo … He won again in the big leagues, alongside the best laboratories in the world in artificial intelligence. How ? We go up the wire.
Confused communication, failed demo … difficult beginnings
November 2022: Chatgpt arrives and changes the situation. In a few days, the Openai tool crosses the milestone of the million users and becomes the general public application to the fastest growth in history. Alphabet, however, had all the assets to dominate the AI market: architecture Transformers was born in its laboratories, and Deepmind constitutes one of the most advanced IA research centers in the world. So facing the Chatgpt technological bomb, Sergey Brin, the co -founder, returns to orchestrate the response even though a “red code” is decreed internally.
“In fact, they realized that they have short-circuited on a market which they were convinced that they were going to be leaders,” analyzes today Clément David, CEO of Theodo Data & AI and Theodo Cloud, companies specializing in the modernization of IT infrastructure and the Cloud. But for the sector specialist, the situation of the time results more from a perception problem than a real technological delay. The failed demonstration of Bard in February 2023, which provides an erroneous response during its public presentation, tarnished the image of the group at the critical moment. More fundamentally, Google suffers from a rough communication: between Vertex, Lamda, Palm and the first versions of Bard, the general public and technological decision -makers are struggling to identify a coherent strategy.
The return to the front of the stage with Gemini
Faced with this crisis of authority over AI, the Mountain View giant quickly operated several levers. The most important thing remains the merger, in April 2023, of Deepmind with Google Brain to give birth to Google Deepmind. A unification of research and development which then aims to accelerate the transformation of scientific advances into concrete commercial products. “It’s for me the time of overthrow,” said Clément David. Beyond the merger, the Deepmind teams are ordered to develop products more turned towards the operational than towards basic research, as confirmed to us by an old frame of Silicon Valley, close to Deepmind.
Reorientation to Business-ST products quickly paid off with the launch of Gemini, Google’s first real commercial success in AI. “The first thing I note is the creation of a strong brand with Gemini,” said Clément David. Unlike the confusion that reigned around the galaxy of previous products, Gemini finally gives a clear and unified identity to the IA offer of Google. A branding strategy, coupled with the technical performance of the model in the benchmarks which makes it possible to rekindle the authority of the brand in the eyes of the general public.
Google, in the head trio since 2024
But the most spectacular demonstration of force remains that of Veo 3, the latest video generation model of Google, which marked the spirits during the Google I/O 2025. A breakthrough which far exceeds technical considerations. “The generation of videos is a bit like your technological showcase. Even if it is not necessarily a market on which you earn money, this is what says if you have your place at the table of big or not,” analyzes Clément David. The stake exceeded the simple technical feat. The multiple comparisons between Sora of Openai and Veo, highlighting Google’s superiority, allowed the Mountain View giant to drive the nail and further credible its know-how.
Beyond this window, Google Deepmind is now competing with Openai and Anthropic on one of the most profitable verticals of AI: code generation. The performance of Gemini 2.5 Pro make Google a reference of the sector alongside Anthropic and Openai. Even more significant, the pace of innovation has accelerated considerably. There is no longer a week without updates, even minor, being deployed throughout the range of IA products from Google.
This new dynamic place Google in a privileged position. Unlike AWS, which develops its own models with Nova but remains dependent on Anthropic for its most advanced solutions, Google fully controls its technological stack. The advantage is even clearer against Microsoft, whose dependence on Openai becomes problematic while tensions between the two partners intensify. For Clément David, this vertical integration is now the master asset of Google: “They have everything internally. They don’t need to buy start-ups.” A decisive advantage that could well make Google the only cloud provider capable of competing technologically with Pure Players of AI, while having the infrastructure necessary to deploy its large -scale models.
